1. Field of Invention
The present invention relates to a CMOS sensor and its method of manufacture. More particularly, the present invention relates to a method of manufacturing a CMOS sensor having a P-I-N structure on a silicon-on-insulator (SOI) substrate
2. Description of Related Art
Charge coupled devices (CCDs) are often employed in digital sensors for image extraction. CCD applications include close-circuit TVs, cameras and video recorders. However, CCDs are quite costly to produce and bulky. Hence, in order to reduce volume, energy consumption and cost, CMOS photo diodes that can be formed by semiconductor techniques are a major substitute for CCDs in the future. At present, CMOS photo diodes have been used inside PC cameras and digital cameras.
A photo diode is a light-sensitive (or light-detecting) semiconductor device having a depletion region with an electric field capable of converting light energy into an electrical signal. When light rays impinge upon the depletion region, atoms within the depletion region are be activated to generate electron-hole pairs. Due to the present of a high electric field within the depletion region, electrons and holes are separated from each other. The electrons migrate towards the N-doped region while the holes migrate towards the P-doped region, thereby leading to a current flow in the intrinsic depletion region. Ideally, the photo diode should be in an open-circuit state without any electric current flow when the device is in total darkness.
As transmission speed requirements continue to increase, response time of photo diodes must also increase correspondingly. Hence, a very thin depletion layer must be maintained inside the photo diode so that transfer time is shortened. However, in order to increase quantum efficiency (that is, the number of electron-hole pairs produced by a given photon), the depletion region must be thick enough to absorb most of the incoming light. Therefore, there must be a trade-off between the thickness of the depletion layer and the response time.
Conventional photo diodes have a P-N photodiode structure. Since the depletion region is a P-N junction formed by the cross-diffusion at the interface of P-doped region and N-doped region, the thickness of the depletion region is difficult to control. Hence, after a voltage is applied to the device, the electric field within the depletion region may not be uniformly distributed.
Another type of photo diode has a P-I-N structure. Because the thickness of the intrinsic depletion region can be adjusted to secure an optimal quantum efficiency and response time, the P-I-N photo diode is one of the most commonly used light-detecting sensors.
FIG. 1 is a schematic, cross-sectional view of a conventional CMOS sensor having a stacked P-I-N structure. The P-I-N structure 102 is formed over a substrate 100. First, an epitaxial method is used to form an N-type region 104, an intrinsic depletion region 106 and a P-type region 108 in sequence on and above the substrate 100. The N-type region 104 and the P-type region 108 can be made with group III-V materials such as GaAs/AlGaAs. The N-type region 104, the intrinsic depletion region 106 and the P-type region 108 together constitute a P-I-N structure 102. A major defect of this type of manufacturing method is the high cost of performing epitaxial growth and the high cost of the group III-V materials.
To reduce the cost of production, conventional semiconductor manufacturing techniques are now employed to form a P-I-N structure. FIG. 2 is a schematic, cross-sectional view of a conventional CMOS sensor having a horizontal P-I-N structure fabricated using common semiconductor techniques. The P-I-N structure 202 is formed by performing ion implantation. P-type ions and N-type ions are implanted into different parts of a substrate 200. The two types of ions are horizontally distributed within the substrate 200, but separated from each other by a distance. An intrinsic depletion region 206 is formed between the P-doped region 208 and the N-doped region 204. The P-type region 208, the intrinsic depletion region 206 and the N-type region 204 together form the P-I-N structure 202. This type of manufacturing method, however, has difficulties in dimensional reduction, especially when a higher level of device integration is demanded. If dimensions of the photo diode are reduced, the intrinsic depletion region 206 must shrink as well. This decreases the light-sensitive area leading to a lower sensitivity.
To resolve device integration problems, another type of conventional CMOS sensor is produced. FIG. 3 is a schematic, cross-sectional view of a conventional CMOS sensor having vertical P-I-N structure embedded within a substrate. The P-I-N structure 302 is formed by performing ion implantation. First, N-type ions are implanted into the substrate 300, and then P-type ions are similarly implanted into the substrate 300 in the same location. Implantation energy level is carefully controlled so that a vertical distribution of ions as shown in FIG. 3 appears. Between the N-type and the P-type regions, an intrinsic depletion region 306 is formed. The N-type region 304, the intrinsic depletion region 306 and the P-type region 308 together form the P-I-N structure 302. Although this method is capable of producing a higher level of integration, size of the depletion region 306 is difficult to control. In other words, the patch of area in the sensitivity region is quite variable.
Furthermore, conventional CMOS sensors generally have the defect of having a short effective interaction length between incoming light and the intrinsic depletion region. Hence, the contrast ratio of the photo diodes is rather low. Taking the P-I-N structure in FIG. 3 as an example, when the incoming light passes into the intrinsic depletion region 306, a portion of the atoms within the region are be activated to create electron-hole pairs. Thereafter, an electric field within the depletion region forces the electrons and holes apart, converting light energy into electrical signals. However, the incoming light interacts with atoms in the depletion region 306 only once, and thus the effective interaction length is short. Consequently, the electrical signal generated by light activation is small, and hence the contrast ratio or sensitivity is low. Moreover, the incoming light is capable of passing through the depletion region 306 to reach the substrate 300. When the substrate 300 absorbs this stray light, unwanted current may be created leading to undesirable video effects.
In light of the foregoing, there is a need to provide an improved CMOS sensor structure and method of manufacture.